WINETASTER ON 03/04/13 WITH 8 JUDGES AND 4 WINES BASED ON RANKS, IDENT=N
Copyright (c) 1995-2013 Richard E. Quandt, V. 1.65
Number of Judges = 8
Number of Wines = 4
Identification of the Wine: The judges' overall ranking:
Wine A is Ch. Gruaud-Larose 1983 ........ 4th place
Wine B is Ch. Haut Brion 1983 ........ 1st place
Wine C is Ch. Ducru Beaucaillou 1983 tied for 2nd place
Wine D is Ch. Lanessan 1983 tied for 2nd place
The Judges's Rankings
Judge Wine -> A B C D
Burt 4. 3. 1. 2.
Ed 1. 3. 2. 4.
Zaki 4. 3. 2. 1.
Orley 2. 1. 4. 3.
Bob 3. 1. 2. 4.
Dilip 4. 3. 1. 2.
Mike 3. 2. 4. 1.
Dick 2. 1. 4. 3.
Table of Votes Against
Wine -> A B C D
Group Ranking -> 4 1 2 2
Votes Against -> 23 17 20 20
( 8 is the best possible, 32 is the worst)
Here is a measure of the correlation in the preferences of the judges which
ranges between 1.0 (perfect correlation) and 0.0 (no correlation):
W = 0.0562
The probability that random chance could be responsible for this correlation
is rather large, 0.7173. Most analysts would say that unless this
probability is less than 0.1, the judges' preferences are not strongly
We now analyze how each taster's preferences are correlated with the group
preference. A correlation of 1.0 means that the taster's preferences are a
perfect predictor of the group's preferences. A 0.0 means no correlation,
while a -1.0 means that the taster has the reverse ranking of the group.
This is measured by the correlation R.
Correlation Between the Ranks of
Each Person With the Average Ranking of Others
Name of Person Correlation R
The wines were preferred by the judges in the following order. When the
preferences of the judges are strong enough to permit meaningful differentiation
among the wines, they are separated by -------------------- and are judged to be
1. ........ 1st place Wine B is Ch. Haut Brion 1983
2. tied for 2nd place Wine C is Ch. Ducru Beaucaillou 1983
3. tied for 2nd place Wine D is Ch. Lanessan 1983
4. ........ 4th place Wine A is Ch. Gruaud-Larose 1983
We now test whether the ranksums AS A WHOLE provide a significant ordering.
The Friedman Chi-square value is 1.3500. The probability that this could
happen by chance is 0.7173
We now undertake a more detailed examination of the pair-wise rank correla-
tions that exist between pairs of judges. First, we present a table in which you
can find the correlation for any pair of judges, by finding one of the names in the
left hand margin and the other name on top of a column. A second table arranges
these correlations in descending order and marks which is significantly positive
significantly negative, or not significant. This may allow you to find clusters
of judges whose rankings were particularly similar or particularly dissimilar.
Pairwise Rank Correlations
Correlations must exceed in absolute value 1.00 for significance at the 0.05
level and must exceed 1.00 for significance at the 0.1 level
Burt Ed Zaki
Burt 1.000 -0.400 0.800
Ed -0.400 1.000 -0.800
Zaki 0.800 -0.800 1.000
Orley -0.800 0.000 -0.600
Bob 0.000 0.200 -0.400
Dilip 1.000 -0.400 0.800
Mike -0.200 -0.800 0.400
Dick -0.800 0.000 -0.600
Orley Bob Dilip
Burt -0.800 0.000 1.000
Ed 0.000 0.200 -0.400
Zaki -0.600 -0.400 0.800
Orley 1.000 0.400 -0.800
Bob 0.400 1.000 0.000
Dilip -0.800 0.000 1.000
Mike 0.400 -0.400 -0.200
Dick 1.000 0.400 -0.800
Burt -0.200 -0.800
Ed -0.800 0.000
Zaki 0.400 -0.600
Orley 0.400 1.000
Bob -0.400 0.400
Dilip -0.200 -0.800
Mike 1.000 0.400
Dick 0.400 1.000
Pairwise correlations in descending order
1.000 Orley and Dick Significantly positive
1.000 Burt and Dilip Significantly positive
0.800 Zaki and Dilip Not significant
0.800 But and Zaki Not significant
0.400 Bob and Dick Not significant
0.400 Zaki and Mike Not significant
0.400 Orley and Mike Not significant
0.400 Mike and Dick Not significant
0.400 Oley and Bob Not significant
0.200 Ed and Bob Not significant
0.000 Burt and Bob Not significant
0.000 Ed and Dick Not significant
0.000 Ed and Orley Not significant
0.000 Bob and Dilip Not significant
-0.200 Burt and Mike Not significant
-0.200 Dilip and Mike Not significant
-0.400 Zaki and Bob Not significant
-0.400 But and Ed Not significant
-0.400 Ed and Dilip Not significant
-0.400 Bob and Mike Not significant
-0.600 Zaki and Orley Not significant
-0.600 Zaki and Dick Not significant
-0.800 Orley and Dilip Not significant
-0.800 But and Orley Not significant
-0.800 Ed and Zaki Not significant
-0.800 Ed and Mike Not significant
-0.800 Dilip and Dick Not significant
-0.800 Burt and Dick Not significant
The goal was to serve four wines, in magnum portions — but poured from bottles —
to see how chateau rankings coordinate with tasting notes. The wines were selected to represent
the quality spectrum from first growth to unclassed growth (albeit one of our host's favorites
for long aging).
In order of their general classification, the Haut Brion sells around $300, the Ducru about $100–
$125, the Gruaud Larose in the same range and the Lanessan about $100. The wines all sell for about
10 times their retail prices in 1986, and have been in our host's cellar since that time.
The wines were very, very close to each other, despite their disparity in price.
Their age appears to have brought these very different growths closer
together, and in particular, the Lanessan showed on a par with the
great Haut Brion. The numbers of wines tasted has nothing to do with
the degree of correlation. All the wines are drinking well, there is no hurry
to finish them now and no harm drinking them now. Contrary to our
normal practice, we tried to identify the wines. Two of the tasters
correctly identified all four wines, but disagreed on the ranking of the
Haut Brion. Both of those tasters agreed that A and C were the Ducru and
the Gruaud Larose which had more in common than B, the Haut Brion.
One taster thought that A, B and D were very elegant and somewhat similar
and C was more lively asnd interesting.
Return to previous page